There are nowhere near 70k engineers working on this - I'd be surprised if even 7k - but this has been in development for years, and will be in development for a few more.
Depending on how many headsets they planned to deliver - I wouldn't be surprised if they each cost more than $1k - there are like 60k infantry soldiers - so that could easily cost >$60M.
I'd imagine they spent >$100M lobbying on this and Jedi.
I can't see how this would cost anywhere near $22Bn - but military contractors are known to have profit margins >30%.
I'm obviously missing a lot of things - but I wouldn't be surprised if this cost MSFT >$10Bn.
Based on what data? Contract bidders must show all direct and indirect costs and are typically allowed to price in ~10% profit when submitting their proposal. In fact, you can look at historical financial data for LMT, NOC, GD, etc and see profit margin float around 7-10%. A proposal with 30% profit margin baked in would be laughed out of the selection committee.
Microsoft could get better engineers for cheaper. Including within the US. The problem is that this isn't about engineering. It's a scheme.
They have become so incompetent that they don't even have the talent necessary to know what talent looks like when recruiting.
Corporate engineers will never hire other engineers who are better than them (for political reasons), so the engineering talent in these companies has been degrading for years as the bureaucracy and internal politics has increased.
Not all experience transfers over and developing completely new technology (both hardware and software in combination) is extremely hard. I think you're vastly underestimating the value people get from these university degrees when they're actually pioneering new things and actually learned the topics they were taught.
Do you work at MSFT? How do you know so much about their competence?
One reason why they suck at web is because they suck at interoperability. They can't do open source properly. They don't know how to optimize software for maximum compatibility. This is because vendor lock-in is the enemy of compatibility and interoperability. Vendor lock-in is in their DNA; everything they do is about lock-in. Their software is all about complex, rigid interfaces; this is how they achieve lock-in but it makes it difficult to integrate their solutions into third party software (high friction, high entry barriers) and difficult to maintain/update.
VScode?
- I'd be very surprised if Apple releases their AR headset without a similar-size (70% or more) investment over the preceding decade. RDT&E for environment-aware devices is incredibly expensive.
- Defense funding and procurement amounts often shock those not in the industry, but the truth is that the entire market for most of what the DoD funds is...the DoD. Sure, a lot of the components and IP can be reused in the commercial sector, but for most programs the DoD has to fund both the initial R&D and the unit sales, because the market is so much smaller and device sales are heavily restricted (e.g. ITAR). The DoD also has MUCH broader and stricter requirements for security, safety, reliability, and compliance with a daunting array of standards and processes than the commercial industry. To be clear, this is on net a good thing for the DoD and by extension the rest of the USA, but it does increase development costs.
Certainly there is profit being made by Microsoft here, that's kind of the point of partnering with the DoD. But the intuitive comparison of "DoD programs should cost the same as developing a commercial equivalent" that many of us make is fundamentally apples and oranges in both the product development and the economics.
If you consider an alternative, just make an app and stick phone on a rifle, sure, that will work for 90% of the stuff you want to do, but it would break and that's bad. A true "heads up display" that you can bet your life on, anywhere on earth, at any time, is just a truly massive undertaking.
It's half as much as the development of the entire flipping Space Shuttle. It's more than Apple's entire yearly R&D budget. It's more than the James Webb Space Telescope and the entire Hubble program (including operations and the repairs that were needed post-launch) combined. It's roughly the cost of 8 Mars rovers of the same type as Curiosity / Perseverance.
This is worse, even, than the Joint Strike Fighter program, which at least started out reasonably with only a pair of 1 billion dollar grants handed out to Boeing and Lockheed to develop two functional prototype fighter aircraft (never mind how the costs spiraled later).
Setting aside the question of how much it should or shouldn't cost to develop such a product, it's absolute insanity to do it with a single massive project. The DoD loves to claim in public they're trying to be more agile, and this is the exact opposite of that. This ought to have been done $100 million (or less) at a time with limited initial scope and incrementally expanding goals.
Don't make excuses for this. There is no possible justification.
For reference, an ounce of gold in the peak at 1980 (which was a long while peak btw) was $800ish, an ounce of gold today is $1800. So more than 2x.
So $22B today is more like $10B in 1980.
Which is around 49 billion in 2020 dollars, adjusted for inflation (gold is a poor comparison, it's closer to 5x than 2x).
At some point you have to make a cost-benefit analysis and look at the bigger picture. For example and in this particular case, yes, spending let’s say 10 billion instead of 22 billion for a rougher solution might increase the dead count on the battlefield in the near future, but if those 12 billion that you’ve now saved go to better free education at home (more engineers in the future to design better weapons) or/and to free healthcare (more fit soldiers in the future) maybe it will all even out at a higher threshold.
Given that soldiers are currently getting by with something that works 0% of the time, the assumption that 90% isn't good enough seems highly suspect. Exactly how many more soldiers will be saved by developing a true heads up display versus an off the shelf app running on a ruggedized phone?
This seems like an obvious case of a failure to actually define requirements and make real engineering decisions.
Do you ask why space travel is expensive? It's arguably much simpler than this is.
If a pair of goggles fails, the soldier takes them off.
These are the same people who equipped soldiers in Vietnam with a rifle that jams due to mud. The idea that military hardware has to be many times more reliable than space hardware is absurd.
Either way, if this is worth doing -- and I agree with you that it's a big, big if -- it is worth doing right.
Increased situational awareness has value, but it is a clear case of diminishing returns. Exactly what the cutoff ought to be is debatable, but that the cutoff exists is certain.
I don't really mean to suggest 22b is too low a cash prize, but I imagine it takes a lot to overcome the barriers that have been wilfully erected. Much less the ones discreetly erected by incumbent military contractors.